Scientists from Google DeepMind and Google Research have open-sourced an AI weather model designed to improve early tropical cyclone warnings, allowing forecasters to predict both a storm’s track and intensity with a single system.
WeatherNext was created with contributions from the National Hurricane Center, the Cooperative Institute for Research in the Atmosphere, the UK Met Office, and other weather agencies. Google has made both the code and model weights available on GitHub to encourage researchers to build on the project.
Google published the WeatherNext research in Nature and summarized the work in a blog post. The researchers explained that tropical cyclones, whether called hurricanes or typhoons, are difficult to forecast because predicting a storm’s path and its intensity typically requires different modeling approaches. WeatherNext was trained to perform both tasks using nearly 20 terabytes of atmospheric data and historical information from the International Best Track Archive for Climate Stewardship.
The researchers said WeatherNext can produce a 15-day forecast in under a minute using a TPU, allowing forecasters to quickly assess the likelihood of severe, high-impact weather events.
Google introduced the second generation of WeatherNext last year. The company is also using AI to improve forecasts for other weather hazards, including flash floods.
Maybe you would like other interesting articles?